Instructions to use nvidia/gpt-oss-puzzle-88B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nvidia/gpt-oss-puzzle-88B with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="nvidia/gpt-oss-puzzle-88B", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("nvidia/gpt-oss-puzzle-88B", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use nvidia/gpt-oss-puzzle-88B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nvidia/gpt-oss-puzzle-88B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nvidia/gpt-oss-puzzle-88B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nvidia/gpt-oss-puzzle-88B

SGLang

How to use nvidia/gpt-oss-puzzle-88B with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "nvidia/gpt-oss-puzzle-88B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nvidia/gpt-oss-puzzle-88B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "nvidia/gpt-oss-puzzle-88B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nvidia/gpt-oss-puzzle-88B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use nvidia/gpt-oss-puzzle-88B with Docker Model Runner:
```
docker model run hf.co/nvidia/gpt-oss-puzzle-88B
```

Fails to run on my Spark with the Docker command from the documentation

by VladNC - opened Mar 27

Discussion

VladNC

Mar 27

First I had to reduce the memory to 0.90, which made it not fail immediately. But now it fails after loading the model, with:

(APIServer pid=532)   File "/usr/local/lib/python3.12/dist-packages/vllm/entrypoints/openai/parser/harmony_utils.py", line 63, in get_encoding
(APIServer pid=532)     _harmony_encoding = load_harmony_encoding(HarmonyEncodingName.HARMONY_GPT_OSS)
(APIServer pid=532)                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(APIServer pid=532)   File "/usr/local/lib/python3.12/dist-packages/openai_harmony/__init__.py", line 699, in load_harmony_encoding
(APIServer pid=532)     inner: _PyHarmonyEncoding = _load_harmony_encoding(name)
(APIServer pid=532)                                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(APIServer pid=532) openai_harmony.HarmonyError: error downloading or loading vocab file: failed to download or load vocab file

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment